int main(int argc, char **argv)
{
if(argc==1) PrintHelp();
scaledCost = 0;
time_t startTime;
srand(time(&startTime));
outputFilename = (char *)"output.txt";
functionMapping = (char *)"UABCDEFGHIJKLMNOPQRSTVWXYZ";//"UEGMPTBFOARDC";
sizeCutoff = 1; densityCutoff=0; pCutoff=1.000001;
ReadParameters(argc, argv);
printf("Graph filename: %s\n",graphFilename);
printf("Number of nodes: %d\n\n",numVerts = ReadClusteringNumber(graphFilename));
if(numVerts== -1){
fprintf(stderr,"Graph input error. File %s is nonexistent, unreadable, or not a valid graph file. QUITTING.\n",graphFilename);
printUsageError();
return 0;
}
printf("Clustering filename: %s\n",clusteringFilename);
printf("Number of clusters: %d\n\n",numClust = ReadClusteringNumber(clusteringFilename));
if(numClust == -1){
fprintf(stderr,"Clustering input error. File %s is nonexistent, unreadable, or not a valid clustering file. QUITTING.\n",clusteringFilename);
printUsageError();
return 0;
}
//printf("Number of complexes: %d\n",numberOfComplexes = ReadClusteringNumber(complexFilename));
printf("Protein name filename: %s\n\n",nameFilename);
NumFunctions = strlen(functionMapping);
printf("%-3dfunctional groups: %s\n\n",NumFunctions, functionMapping);
printf("Prediction cutoffs:\n Size >= %d.\n Density >= %-1.3f.\n P-value <= %-.1e.\n\n",
sizeCutoff, densityCutoff, pCutoff);
/* FIX COMPUTATION OF BEST P-VALUE.
* WRITE THE REST OF THE CHAPTER.*/
//initialize the arrays.
ComplexList = new SLList[numberOfComplexes];
proteinName = new std::string[numVerts];
Function = new char[numVerts];
FunctionNum = new int[numVerts];
FunctionSize = new int[NumFunctions];
GetProteinNames(nameFilename);
whichCluster = new int[numVerts];
AlreadyCounted = new bool[numVerts];
ComplexSize = new int[numberOfComplexes];
numberMatched = new int[numberOfComplexes];
numClust = CountClusters(clusteringFilename);
ClusterSize = new int[numClust];
ClusterDensity = new float[numClust];
ComplexDensity = new float[numberOfComplexes];
ClusterP = new double[numClust];
ComplexP = new double[numberOfComplexes];
ComplexPee = new double[numberOfComplexes];
ClusterFunction = new int[numClust];
ComplexFunction = new int[numberOfComplexes];
ClusterMatched = new bool[numClust];
ComplexMatched = new bool[numberOfComplexes];
ClusterSMatched = new bool[numClust];
ComplexSMatched = new bool[numberOfComplexes];
MainInComplex = new int[numberOfComplexes];
MainInCluster = new int[numClust];
UnknownInComplex = new int[numberOfComplexes];
UnknownInCluster = new int[numClust];
ComplexList = new SLList[numberOfComplexes];
MaxMatch = new int[numClust];
MaxMatchPct = new float[numClust];
MaxMatchWith = new int[numClust];
MaxMatchPctWith = new int[numClust];
for(int clust = 1; clust < numClust; clust++){
MaxMatch[clust]=0;
MaxMatchPct[clust]=0;
MaxMatchWith[clust]=0;
MaxMatchPctWith[clust]=0;
}
ReadClustering(clusteringFilename);
graph.SetNumClust(numClust);
//printf("%d clusters....\n",(int) graph.NUM_CLUST);
graph.MakeGraph(graphFilename);
graph.CheckAdjList();
graph.FillAdjList();
graph.InitClustering(whichCluster);
//graph.SetInitialNumAndDom();
//Matching criteria
//ComplexCutoff=.7; ClusterCutoff=.7;
//ComConCutoff=1.1; CluConCutoff=.9;
//ReadComplexes(complexFilename);
// GenVertexPermutation(graph.Order);
//GetComplexP();
GetClusterP();
// Match();
GetStats();
//WriteData(outputFilename);
printf("\n");
WriteVerbalData2(outputFilename);
//printf("PARAMS: %d %f %f\n", sizeCutoff,densityCutoff, pCutoff);
//WriteComposition(outputFilename);
/*
for(int clust=0; clust<numClust; clust++){
if(!ClusterMatched[clust]){
printf("Cluster %d unmatched. Tops are %d/%d with %d (size %d) and %f with %d (size %d).\n",
clust, MaxMatch[clust], (int) graph.ClusterSize[clust],
MaxMatchWith[clust], ComplexSize[MaxMatchWith[clust]],
MaxMatchPct[clust], MaxMatchPctWith[clust],
ComplexSize[MaxMatchPctWith[clust]]);
}
}
*/
return 0;
}
// RunExperiment performs a single experiment of RNSC. See Fig 3.1 in my
// Master's thesis for an explanation.
void Experiment::RunExperiment(Graph& graph)
{
if(!P_SkipNaive){//Run the naive scheme.
if(SILENCE<=2) printf("\tRunning naive scheme....\n");
//Initialize the clustering C_0 for the experiment.
graph.InitClustering();
//Initialize the naive move scheme, i.e. MoveList, NaiveCost,
//FirstWithCost, NumWithCost, MoveListPtr.
graph.InitNaiveScheme();
//Set the best clustering to the initial clustering.
BestNaiveCost = graph.NaiveCost;
for(int i=0; i<graph.Order; i++){
ClusteringS[i] = graph.WhichCluster[i];
}
//Run the naive scheme.
if(SILENCE<=1) printf("\t\tInitial naive cost \t=%14d\n",BestNaiveCost);
RunNaiveScheme(graph);
if(SILENCE<=1) printf("\t\tFinal naive cost \t=%14d\n", BestNaiveCost);
//Delete the move list.
graph.PurgeMoveList();
//if(SILENCE<=1) printf("%d\t",graph.LoadClustering(ClusteringS));
//Adjust the clustering, i.e. set the graph's current clustering to C_n.
graph.InitClustering(ClusteringS);
} else {//the naive scheme was not run.
//We need to initialize the clustering C_0 for the experiment.
//That is, we don't run the naive scheme, so C_n = C_0.
graph.InitClustering();
}
if(P_ScaledStoppingTol > 0){//Run the scaled scheme.
if(SILENCE<=2) printf("\tRunning scaled scheme....\n");
//Initialize the scaled move scheme, i.e. MoveList, ScaledCost,
//FirstWithCost, NumWithCost, MoveListPtr.
graph.InitScaledScheme();
//Set the best clustering to the initial clustering, i.e. C_s := C_n.
BestScaledCost = graph.ScaledCost;
for(int i=0; i<graph.Order; i++){
ClusteringS[i] = graph.WhichCluster[i];
}
//Run the scaled scheme.
if(SILENCE<=1) printf("\t\tInitial scaled cost \t= %13.3f\n", BestScaledCost);
RunScaledScheme(graph);
if(SILENCE<=1) printf("\t\tFinal scaled cost \t= %13.3f\n", BestScaledCost);
//Clean up the scaled move scheme.
graph.PurgeMoveList();
//if(SILENCE<=1) printf(" %d\n",graph.LoadClustering(ClusteringS));
//Adjust the clustering, i.e. set the graph's current clustering to C_s.
graph.InitClustering(ClusteringS);
} else {//The scaled scheme is not run. Get the scaled cost for C_n.
BestScaledCost = graph.GetScaledCostForClustering();
}
}